ACR Bulletin

Covering topics relevant to the practice of radiology

Marrying QI and Informatics for Better Patient Care

Patient safety and operational efficiency hinge on data-driven strategies to support quality.
Jump to Article

"Radiology is way ahead of other specialties in FDA-approved AI and machine learning devices."

—Keith J. Dreyer, DO, PhD, FACR, ACR DSI chief science officer
December 01, 2023

Quality improvement (QI) and informatics thought leaders came together in October in San Diego for the ACR 2023 Quality & Safety + Informatics Conference, a special event showcasing the proven and potential benefits of a symbiotic approach to improving patient outcomes and advancing operational practices.

QI and quality assurance initiatives involve studying a problem, identifying root causes and key drivers, and conducting tests of change for process improvement. In tandem, informatics supports reliable workflows, drives efficiency and reduces human error.

“Quality and safety (Q&S) and informatics are two core components of operations in any organization,” says Mythreyi B. Chatfield, PhD, ACR executive vice president for quality and safety. “Q&S identifies system-level problems and focuses on change management — but changes are not scalable or sustainable without informatics. Q&S identifies the ‘what’ and ‘why’ of the important challenges we face today. Informatics answers the ‘how’ and allows us to imagine the future." The ACR’s Commission on Informatics works to advance the College’s data systems and interfaces, engage with stakeholders and deliver content to radiologists through integrated systems.

“The QS+I conference presented a unique collaboration between informatics and Q&S leaders to encourage discussions that cannot happen in a vacuum,” says Mike Tilkin, MS, ACR executive vice president and chief information officer. “By creating the context and opportunity to explore both the current reality in both disciplines as well as the art of the possible, the gathering explored the potential to identify game-changing strategies with an ultimate goal of improving patient care.”

The Evolution of Quality

Keynote speaker Kedar S. Mate, MD, president and CEO of the Institute for Healthcare Improvement (IHI) and a faculty member at Weill Cornell Medical College, described the evolution of quality as creating a healthcare environment in which you are constantly working with patients and their families in the community — to become a service organization, not just a healthcare facility or health system. 

“When patients work with their healthcare providers, outcomes are better,” Mate told the audience of radiologists, RTs and other staff focused on quality. “The co-production of care is centered on trust and providing patients with the best information in real time.” 

People plus operating systems creates incredible results and fosters a culture that values equity and joy in the workplace, Mate said.

Treating the right patients with the right test at the right time and at the right price should be the mantra of all QI activities, said presenter Ruth C. Carlos, MD, MS, FACR, professor and assistant chair of clinical research in the department of radiology at the University of Michigan and JACR® editor-in-chief. Carlos presented the business case for health equity, stressing that radiologists should be “doing well by doing good.”

There are disparities in access to screening, for example, and high out-of-pocket costs can hinder adherence to screening recommendations. Through legislation, some states have been working to prohibit insurers from imposing copayments, coinsurance or deductibles for supplemental screening, she told attendees. 

Hopefully the trend will continue, she said, with increased state and national mandates to improve health equity. “Increased investment to improve care in minority populations has the potential to increase total revenue from increased service delivery,” Carlos said. “And future Medicare alternative payment models — which include incentive payments for high-quality care — will likely include health equity metrics.”

Employing financial navigators can mitigate financial losses for both patients and health systems, said Gelareh Sadigh, MD, director of health services and comparative outcome research at University of California Irvine. Sadigh talked about financial toxicity, when conditions arise from out-of-pocket medical expenses and decreases in income due to the inability to work during or following treatment. 

“The majority of patients want to know about price before they receive medical services,” Sadigh said. “Financial hardship is a big health equity issue, and screening for such hardships should be implemented as part of clinical visits.” Financial education, screening and price transparency are potential solutions to mitigate financial hardship, she said, and seemingly small initiatives add up. A rideshare program, for example, may cost just hundreds of dollars per patient while no-shows cost thousands of dollars, Sadigh illustrated. “The cost of transportation is nothing compared to missed appointments and could save millions, especially in rural areas.” 

Rural populations have a higher mortality rate than urban populations. There are more deaths in rural areas from cancer as well as heart disease, stroke, kidney disease, respiratory disease, diabetes, unintentional injuries and even suicide. These statistics make a strong business case for health equity in rural populations, said Anand Narayan, MD, PhD, associate professor and vice chair of equity, and associate director of diversity, equity and inclusion at the Carbone Cancer Center at University of Wisconsin Madison.

“The economic burden of healthcare inequity is comprised of higher medical care expenditures, lost labor market productivity and premature deaths,” Narayan said. The costs to treat cancer as it progresses, for example, rise substantially. 

There is a need to leverage social determinants of health to improve screening rates. “Access to radiologists is a problem,” he said, “and ER visits cost more than seeing your primary care physician.”

A continued need for more radiologists in rural areas with disadvantaged populations worsens the problem, Narayan said. Safety net hospitals serving Medicaid patients are also struggling to keep up, another presenter noted. They tend to operate on thin profit margins and depend on public funding to serve vulnerable populations.

Other Q&S topics included the success and future expansion of the ACR’s Learning Network, inclusive patient consent, clinical decision support, more patient-friendly radiology reports, the importance of measuring patient satisfaction and improving wait times for imaging results. The informatics group spoke to the importance of implementing data-driven strategies and vetted AI tools to overcome these and other challenges within the specialty — and medicine in general.

The Value of Good Data and AI in Clinical Use

Healthy data management systems need consistency in data and results. Presenters posed the question, Is your health system healthy? The answer lies in informatics that can guide next steps toward QI improvements and operational effectiveness. As one presenter said, “Don’t assume everything is OK just because your dashboard is green. Take a hard look at your institution’s data management system to find potential improvements.”

As an example, use informatics to investigate how long patients wait to get a screening mammogram and how long they must wait for the results. Ask what changes you could make to your system to improve upon what you find. Presenters used the metric of sleepless nights by patients awaiting results and setting a goal to reduce the time between screening, interpretation, biopsies (if needed) and reporting results to patients. There were breakout sessions on how to identify problems — such as screening outreach and improving adherence to follow-up recommendations — and how to create a process control guide.

As anticipated, the informatics portion of the conference also focused heavily on AI and its current and future uses in radiology and beyond. There was a common theme throughout all AI-related presentations that a human must be in the loop when AI is used for evaluating patient images. There was also an emphasis on including all stakeholders when putting AI tools to use — from referrers to IT staff to patients and radiologists.

With these considerations in mind, there is no question that “the hype is real,” said Tessa S. Cook, MD, PhD, associate professor of radiology and vice chair of practice transformation for the Perelman School of Medicine at the University of Pennsylvania. 

“Radiology is way ahead of other specialties in FDA-approved AI and machine learning devices,” said Keith J. Dreyer, DO, PhD, FACR, ACR DSI chief science officer and vice president of enterprise medical imaging for Mass General Brigham. He added that in the AI space, radiologists and their teams will be constantly reinventing themselves. This is especially true when using generative AI. 

Speakers stressed the importance of safety when using AI tools and spoke about large-scale deployment and quality management of internally developed algorithms versus those created by vendors. Attendees were shown a host of new mind-bending capabilities of generative AI tools — including creating images and video from text using ChatGPT and other generative AI applications. Dreyer noted that most people are way behind in using the latest iterations of such tools and that admittedly it is difficult to keep up.

Another speaker echoed this challenge. “If I don’t read at least three hours a day on AI, I am behind,” said Woojin Kim, MD, chief medical information officer at Rad AI. Kim cautioned that trusting AI findings based on the assumption that what is presented in all literature is accurate or true is ill-advised. You must be mindful, Kim said, with large language models generating impressions because they may be trained on small clinical datasets with a narrow scope. He showed several ChatGPT-generated reports, for example, that were completely wrong and needed extensive review and correction by a radiologist.

The ACR has been working extensively on AI use cases through its Data Science Institute®. There was a presentation on AI Central, which provides detailed information on FDA-cleared AI medical products related to radiology and other imaging domains. Additional AI discussions centered on respecting health information privacy, using AI to assist referrers in decision-making, educating patients and leveraging these new tools to improve health equity.

Recognizing the trepidation of some, David B. Larson, MD, MBA, FACR, professor and vice chair of education and clinical operations in the department of radiology at Stanford University School of Medicine — and chair of the ACR Commission on Quality and Safety — offered a calming analogy. He compared the introduction of AI in radiology to the addition of automatic transmissions in vehicles. 

“At one time, people only knew how to operate cars manually — they had to learn to use something new,” Larson said. “It turned out to be something that actually improved their lives by making driving easier. It’s just something new to embrace.”

Q&S + Informatics 2024

Look for more cutting-edge food for thought at the 2024 event — celebrating collaborative quality improvement and the complementary and transformative tools that will propel the specialty into its next 100 years. Until then, be sure to check out on-demand videos from this year’s event.

Author By Chad E. Hudnall,  senior writer, ACR Press